Comments on: YUI 2.7.0 on TaskSpeed

By: Claude

Claude — Thu, 07 May 2009 23:33:37 +0000

To answer my own question... Tim Sporcic did include Ext-JS Core (beta) in a slickspeed test, take a look.

By: Peter Higgins

Peter Higgins — Tue, 21 Apr 2009 03:37:55 +0000

@Eric – Awesome, just got the update. Thanks for the quick turnaround. You guys have been such a joy to work with on this.

Regards,
Peter

By: Eric Ferraiuolo

Eric Ferraiuolo — Tue, 21 Apr 2009 01:05:43 +0000

I've updated the finale test for YUI 2.7.0 to match the strategy taken by PureDOM: looping over the body's children and removing the Nodes one at a time. I've pushed the new code to my fork on GitHub.

By: Peter Higgins

Peter Higgins — Mon, 20 Apr 2009 14:08:54 +0000

Actually, the Dojo 1.2.3 tests are using _destroyElement in finaly. The YUI test should probably be adjusted to use the PureDom way, as innerHTML=”" won’t cleanup everything is should as the removeChild way (in IE, specifically). Dojo’s empty() uses innerHTML=”" on the fast path (where it knows it can) but uses removeChild where it must.

Regards,
Peter Higgins
Dojo Toolkit Project Lead

By: bruce

bruce — Thu, 16 Apr 2009 05:06:34 +0000

@Eric – No complaints here, I only looked into it because I was curious how YUI got to be faster than pure DOM manipulation. I’d say you did the right thing, and it hardly matters anyway because all of Luke’s caveats about not taking these sorts of tests too literally are correct. Cheers!

By: Claude

Claude — Thu, 16 Apr 2009 04:27:42 +0000

Is there a reason why Ext-JS core was left out of these tests?

By: Eric Ferraiuolo

Eric Ferraiuolo — Tue, 14 Apr 2009 18:13:50 +0000

@bruce When I was writing the finale test for YUI I looked inside Dojo’s empty method and they are setting the node’s innerHTML to “”. Since YUI doesn’t have an empty method, I did the same thing.

By: bruce

bruce — Mon, 13 Apr 2009 23:12:52 +0000

Interesting stuff, and well presented.

A minor quibble with the ‘finale’ test though; apparently the YUI way to clear the page is:

document.body.innerHTML = "";

… but PureDom has to do this:

while (body.firstChild) body.removeChild(body.firstChild);

PureDom got a little shortchanged on that one! The PureDom developers should be on the street protesting with placards, I tell you.

By: Eric Clemmons

Eric Clemmons — Mon, 13 Apr 2009 22:29:22 +0000

Any sort of benchmarking utility should be taken with a grain of salt, but they’re pretty important for certain tasks.

For example, suppose I generate custom charts and graphs for hundreds of terms on a page. If I’m making simple DOM adjustments (removing/adding elements & classes), then TaskSpeed will give some indication on what framework to use for that operation.

On the other hand, if I’m performing complex CSS Select queries on these hundreds of elements, then SlickSpeed will help determine the appropriate framework there.